batchalign 0.7.22.post3__tar.gz → 0.7.22.post5__tar.gz

{batchalign-0.7.22.post3/batchalign.egg-info → batchalign-0.7.22.post5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: batchalign
-Version: 0.7.22.post3
+Version: 0.7.22.post5
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu
@@ -43,6 +43,7 @@ Requires-Dist: praat-parselmouth==0.4.6
 Requires-Dist: opensmile>=2.5.0
 Requires-Dist: pyannote.audio
 Requires-Dist: onnxruntime
+Requires-Dist: certifi>=2025.10.5
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Provides-Extra: train

{batchalign-0.7.22.post3 → batchalign-0.7.22.post5}/batchalign/pipelines/avqi/engine.py

@@ -11,6 +11,7 @@ from typing import Tuple, Dict, Optional
 import os
 from pathlib import Path
 import logging
+import torchaudio
 
 from batchalign.pipelines.base import BatchalignEngine
 from batchalign.document import Task, Document
@@ -246,8 +247,37 @@ class AVQIEngine(BatchalignEngine):
         
         L.info(f"Calculating AVQI for CS: {cs_path.name}, SV: {sv_name}")
 
+        # Create mono versions of both files
+        cs_mono_path = None
+        sv_mono_path = None
+
         try:
-            avqi_score, features = self.calculate_avqi_features(str(cs_file), str(sv_file))
+            # Convert cs_file to mono
+            L.info(f"Converting {cs_path.name} to mono")
+            cs_waveform, cs_sample_rate = torchaudio.load(str(cs_file))
+            if cs_waveform.shape[0] > 1:
+                cs_mono = cs_waveform.mean(dim=0, keepdim=True)
+            else:
+                cs_mono = cs_waveform
+
+            # Create mono filename: file_name.cs.[extension].mono
+            cs_mono_path = cs_path.parent / f"{cs_path.name}.mono.wav"
+            torchaudio.save(str(cs_mono_path), cs_mono, cs_sample_rate)
+
+            # Convert sv_file to mono
+            L.info(f"Converting {sv_name} to mono")
+            sv_waveform, sv_sample_rate = torchaudio.load(str(sv_file))
+            if sv_waveform.shape[0] > 1:
+                sv_mono = sv_waveform.mean(dim=0, keepdim=True)
+            else:
+                sv_mono = sv_waveform
+
+            # Create mono filename: file_name.sv.[extension].mono
+            sv_mono_path = sv_file.parent / f"{sv_file.name}.mono.wav"
+            torchaudio.save(str(sv_mono_path), sv_mono, sv_sample_rate)
+
+            # Calculate AVQI using mono versions
+            avqi_score, features = self.calculate_avqi_features(str(cs_mono_path), str(sv_mono_path))
 
             results = {
                 'avqi': avqi_score,
@@ -277,4 +307,12 @@ class AVQIEngine(BatchalignEngine):
                 'tilt': 0.0,
                 'error': str(e),
                 'success': False
-            }
+            }
+        finally:
+            # Clean up temporary mono files
+            if cs_mono_path and cs_mono_path.exists():
+                L.info(f"Cleaning up temporary file: {cs_mono_path.name}")
+                cs_mono_path.unlink()
+            if sv_mono_path and sv_mono_path.exists():
+                L.info(f"Cleaning up temporary file: {sv_mono_path.name}")
+                sv_mono_path.unlink()

batchalign-0.7.22.post5/batchalign/version

@@ -0,0 +1,3 @@
+0.7.22-post.5
+October 19th, 2025
+Bumped AVQI calculation method

{batchalign-0.7.22.post3 → batchalign-0.7.22.post5/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: batchalign
-Version: 0.7.22.post3
+Version: 0.7.22.post5
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu
@@ -43,6 +43,7 @@ Requires-Dist: praat-parselmouth==0.4.6
 Requires-Dist: opensmile>=2.5.0
 Requires-Dist: pyannote.audio
 Requires-Dist: onnxruntime
+Requires-Dist: certifi>=2025.10.5
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Provides-Extra: train

{batchalign-0.7.22.post3 → batchalign-0.7.22.post5}/batchalign.egg-info/requires.txt

@@ -33,6 +33,7 @@ praat-parselmouth==0.4.6
 opensmile>=2.5.0
 pyannote.audio
 onnxruntime
+certifi>=2025.10.5
 
 [dev]
 pytest

{batchalign-0.7.22.post3 → batchalign-0.7.22.post5}/setup.py

@@ -64,7 +64,8 @@ setup(
         "praat-parselmouth==0.4.6",  
         "opensmile>=2.5.0",
         "pyannote.audio",
-        "onnxruntime"
+        "onnxruntime",
+        "certifi>=2025.10.5"
     ],
     extras_require={
         'dev': [

batchalign-0.7.22.post3/batchalign/version

@@ -1,3 +0,0 @@
-0.7.22-post.3
-October 16th, 2025
-AVQI Patches, etc.